Re: Is it better to set the freelist to a prime number ?

Jonathan Lewis wrote:
>
> I think it is purely down to the choice
> of algorithm. The hash function Oracle
> uses for hash partitions is clearly
> designed as a 2^N algorithm for mechanical
> reasons. But I have a memory that the best
> randomisers always used prime numbers
> for the hash target.
>
> --
> Jonathan Lewis
> http://www.jlcomp.demon.co.uk
>
> Next Seminars
> UK Sept
> Australia August
> Malaysia September
> USA x 2 November
>
> http://www.jlcomp.demon.co.uk/seminar.html
>
> Yong Huang wrote in message
> <07b6be7c6105b142172a9b70371d2d92.99975_at_mygate.mailgate.org>...
> >"Jonathan Lewis" <jonathan_at_jlcomp.demon.co.uk> wrote in message
> >news:1026200922.18484.1.nnrp-13.9e984b29_at_news.demon.co.uk
> >
> >>
> >> When a process requests a free list,
> >> it uses some sort of 'hashing' function
> >> to select which free list, based on the
> >> process id. Using a prime number
> >> with such mechanisms usually reduces
> >> the number of collisions that occur if
> >> the input is randomly distributed.
> >
> >Jonathan, I think you meant non-randomly, or not sufficiently randomly.
> >Otherwise why does the prime number matter? But I could be wrong.
> >
> >Yong Huang
> >
> >
> >--
> >Posted via Mailgate.ORG Server - http://www.Mailgate.ORG

Do we (anyone?) have any insider info on the hash algorithm used for freelist mapping? Is it plain ol' modulo - in which case, the use of prime number would be suitable, but there's also endless (non-modulo) hashing algoriths out there that would not have such a requirement

A nice hash site is http://www.burtleburtle.net/bob/hash/evahash.html

Cheers
Connor

-- 
==============================
Connor McDonald

http://www.oracledba.co.uk

"Some days you're the pigeon, some days you're the statue..."